\begin{table}[!htbp]
\centering  
\caption{\label{TestResultsTradeWar_outcomes} 
 Testing Predictions about the Impact of the US-China Trade War}  
\begin{tabular}{lccc}\toprule

Outcome: \hspace{5.3cm} & $\hspace{0.85cm} \Delta y_n \hspace{0.85cm}$ & $\hspace{0.66cm} \Delta x_n (\hat{\theta}) \hspace{0.64cm}$ & $\Delta y_n - \Delta x_n (\hat{\theta})$ \\ 
& (1) & (2) & (3) \\ \hline

\multicolumn{4}{l}{\emph{Panel A: All outcomes}}\\							
$\quad$Point estimate	&	0.50	&	0.35	&	0.15	\\
$\quad$ Std. error	&	0.22	&	0.02	&	0.23	\\
$\quad$ p-value of $H_0$: $\hat{\beta}=0$	&	0.04	&	0.00	&	0.52	\\ [8pt]
							
							
\multicolumn{4}{l}{\emph{Panel B: Export prices}}\\							
$\quad$Point estimate	&	1.14	&	1.21	&	-0.07	\\
$\quad$ Std. error	&	0.78	&	0.04	&	0.84	\\
$\quad$ p-value of $H_0$: $\hat{\beta}=0$	&	0.16	&	0.00	&	0.93	\\ [8pt]
							
							
\multicolumn{4}{l}{\emph{Panel C: Import prices}}\\							
$\quad$Point estimate	&	0.54	&	0.21	&	0.33	\\
$\quad$ Std. error	&	0.06	&	0.06	&	0.08	\\
$\quad$ p-value of $H_0$: $\hat{\beta}=0$	&	0.00	&	0.00	&	0.00	\\ [8pt]
							
							
\multicolumn{4}{l}{\emph{Panel D: Tariff revenue}}\\							
$\quad$Point estimate	&	1.14	&	0.37	&	0.77	\\
$\quad$ Std. error	&	0.22	&	0.06	&	0.21	\\
$\quad$ p-value of $H_0$: $\hat{\beta}=0$	&	0.00	&	0.00	&	0.00	\\ [6pt]





 \bottomrule
\end{tabular}\\
\vspace{-0.3cm}
\justify \footnotesize \emph{Notes:} Sample of changes in 24,193 welfare-relevant outcomes in Panel A, 5,687 exported varieties with data on prices in Panel B, and 9,253 imported varieties with data on prices and duties in Panels C and D. In the sample $\mathcal{N}$ associated with each panel, we use the preferred IV $z^{\text{pref}}$, as defined by equations (\ref{eq:preferred}) and (\ref{eq:tariff_shifter_application}), to compute: $\frac{1}{|\mathcal{N}|}\sum_{n\in \mathcal{N}} z^{\text{pref}}_n \Delta y_n$ with $\Delta y_n$ the actual change in outcome $n$, in column (1); $\frac{1}{|\mathcal{N}|}\sum_{n\in \mathcal{N}} z^{\text{pref}}_n \Delta x_n (\hat{\theta})$ with $\Delta x_n (\hat{\theta})$ the predicted change in outcome $n$ using FGKK's estimates $\hat{\theta}$ described in Section \ref{subsec:Simulation-Procedure}, in column (2); the IV-based test  $\frac{1}{|\mathcal{N}|}\sum_{n\in \mathcal{N}} z^{\text{pref}}_n (\Delta y_n - \Delta x_n (\hat{\theta}))$, in column (3). Inference accounting for the estimation of $\hat{\theta}$ as described in Appendix \ref{subsec:Implementation}.
\end{table}